Limits...
Canonical source reconstruction for MEG.

Mattout J, Henson RN, Friston KJ - Comput Intell Neurosci (2007)

Bottom Line: Electromagnetic lead fields are computed using the warped mesh, in conjunction with a spherical head model (which does not rely on individual anatomy).This enables the pooling of data from multiple subjects and the reporting of results in stereotactic coordinates.Furthermore, it allows the graceful fusion of fMRI and MEG data within the same anatomical framework.

View Article: PubMed Central - PubMed

Affiliation: INSERM U821, Dynamique Cérébrale et Cognition, Lyon, France. jeremiemattout@yahoo.fr

ABSTRACT
We describe a simple and efficient solution to the problem of reconstructing electromagnetic sources into a canonical or standard anatomical space. Its simplicity rests upon incorporating subject-specific anatomy into the forward model in a way that eschews the need for cortical surface extraction. The forward model starts with a canonical cortical mesh, defined in a standard stereotactic space. The mesh is warped, in a nonlinear fashion, to match the subject's anatomy. This warping is the inverse of the transformation derived from spatial normalization of the subject's structural MRI image, using fully automated procedures that have been established for other imaging modalities. Electromagnetic lead fields are computed using the warped mesh, in conjunction with a spherical head model (which does not rely on individual anatomy). The ensuing forward model is inverted using an empirical Bayesian scheme that we have described previously in several publications. Critically, because anatomical information enters the forward model, there is no need to spatially normalize the reconstructed source activity. In other words, each source, comprising the mesh, has a predetermined and unique anatomical attribution within standard stereotactic space. This enables the pooling of data from multiple subjects and the reporting of results in stereotactic coordinates. Furthermore, it allows the graceful fusion of fMRI and MEG data within the same anatomical framework.

No MeSH data available.


Related in: MedlinePlus

Spatialnormalization scheme.
© Copyright Policy - open-access
Related In: Results  -  Collection


getmorefigures.php?uid=PMC2266807&req=5

fig2: Spatialnormalization scheme.

Mentions: Spatial normalization is a term that refers to thewarping or mapping of a subject specific image into a standard anatomicalspace. It is used routinely in fMRI and PET to enable inter-subject pooling.The parameters θi that define thetransformation x(0) → x(n) are identifiedusing a Bayesian scheme that incorporates constraints on the smoothness of thetransformation [2]. xi(n) represents theposition of the i th controlpoint after n iterations. Inbrief, the warping is parameterized in terms of spatial basis functions (inSPM, we use a discrete cosine set). These encode the change in positioneffected by each transform parameter ∂x/∂θi The coefficients of these basis functions maximize theirconditional probability (i.e., maximize the likelihood and prior density). Thelikelihood is computed using a forward model, which mixes several canonicaltemplates and then warps them to predict the observed image. The mismatchbetween the warped mixture of templates and the observed image constitutes aprediction error. Under Gaussian assumptions this error gives the likelihood ofthe observed image, given the mixing and warping parameters. Roughtransformations are penalized by appropriate shrinkage priors on thecoefficients, formulated in terms of their covariance. The parameters arecomputed using a Newton method. The inverse of the template warping is appliedto the image and the process iterated until convergence and the image isspatially normalized (see Figure 2 for a schematic).


Canonical source reconstruction for MEG.

Mattout J, Henson RN, Friston KJ - Comput Intell Neurosci (2007)

Spatialnormalization scheme.
© Copyright Policy - open-access
Related In: Results  -  Collection

Show All Figures
getmorefigures.php?uid=PMC2266807&req=5

fig2: Spatialnormalization scheme.
Mentions: Spatial normalization is a term that refers to thewarping or mapping of a subject specific image into a standard anatomicalspace. It is used routinely in fMRI and PET to enable inter-subject pooling.The parameters θi that define thetransformation x(0) → x(n) are identifiedusing a Bayesian scheme that incorporates constraints on the smoothness of thetransformation [2]. xi(n) represents theposition of the i th controlpoint after n iterations. Inbrief, the warping is parameterized in terms of spatial basis functions (inSPM, we use a discrete cosine set). These encode the change in positioneffected by each transform parameter ∂x/∂θi The coefficients of these basis functions maximize theirconditional probability (i.e., maximize the likelihood and prior density). Thelikelihood is computed using a forward model, which mixes several canonicaltemplates and then warps them to predict the observed image. The mismatchbetween the warped mixture of templates and the observed image constitutes aprediction error. Under Gaussian assumptions this error gives the likelihood ofthe observed image, given the mixing and warping parameters. Roughtransformations are penalized by appropriate shrinkage priors on thecoefficients, formulated in terms of their covariance. The parameters arecomputed using a Newton method. The inverse of the template warping is appliedto the image and the process iterated until convergence and the image isspatially normalized (see Figure 2 for a schematic).

Bottom Line: Electromagnetic lead fields are computed using the warped mesh, in conjunction with a spherical head model (which does not rely on individual anatomy).This enables the pooling of data from multiple subjects and the reporting of results in stereotactic coordinates.Furthermore, it allows the graceful fusion of fMRI and MEG data within the same anatomical framework.

View Article: PubMed Central - PubMed

Affiliation: INSERM U821, Dynamique Cérébrale et Cognition, Lyon, France. jeremiemattout@yahoo.fr

ABSTRACT
We describe a simple and efficient solution to the problem of reconstructing electromagnetic sources into a canonical or standard anatomical space. Its simplicity rests upon incorporating subject-specific anatomy into the forward model in a way that eschews the need for cortical surface extraction. The forward model starts with a canonical cortical mesh, defined in a standard stereotactic space. The mesh is warped, in a nonlinear fashion, to match the subject's anatomy. This warping is the inverse of the transformation derived from spatial normalization of the subject's structural MRI image, using fully automated procedures that have been established for other imaging modalities. Electromagnetic lead fields are computed using the warped mesh, in conjunction with a spherical head model (which does not rely on individual anatomy). The ensuing forward model is inverted using an empirical Bayesian scheme that we have described previously in several publications. Critically, because anatomical information enters the forward model, there is no need to spatially normalize the reconstructed source activity. In other words, each source, comprising the mesh, has a predetermined and unique anatomical attribution within standard stereotactic space. This enables the pooling of data from multiple subjects and the reporting of results in stereotactic coordinates. Furthermore, it allows the graceful fusion of fMRI and MEG data within the same anatomical framework.

No MeSH data available.


Related in: MedlinePlus